Spaces:
Paused
Paused
File size: 7,469 Bytes
45ee559 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 |
{
"cells": [
{
"cell_type": "markdown",
"metadata": {
"Collapsed": "false"
},
"source": [
"# Jupyter Notbook for phoneme coverage analysis\n",
"\n",
"This jupyter notebook checks dataset configured in config.json for phoneme coverage.\n",
"As mentioned here https://github.com/mozilla/TTS/wiki/Dataset#what-makes-a-good-dataset a good phoneme coverage is recommended.\n",
"\n",
"Most parameters will be taken from config.json file in mozilla tts repo so please ensure it's configured correctly for your dataset.\n",
"This notebook used lots of existring code from the TTS repo to ensure future compatibility.\n",
"\n",
"Many thanks to Neil Stoker supporting me on this topic :-).\n",
"\n",
"I provide this notebook without any warrenty but it's hopefully useful for your dataset analysis.\n",
"\n",
"Happy TTS'ing :-)\n",
"\n",
"Thorsten Müller\n",
"\n",
"* https://github.com/thorstenMueller/deep-learning-german-tts\n",
"* https://discourse.mozilla.org/t/contributing-my-german-voice-for-tts/"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"Collapsed": "false"
},
"outputs": [],
"source": [
"# set some vars\n",
"# TTS_PATH = \"/home/thorsten/___dev/tts/mozilla/TTS\"\n",
"CONFIG_FILE = \"/path/to/config/config.json\"\n",
"CHARS_TO_REMOVE = \".,:!?'\"\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"Collapsed": "false"
},
"outputs": [],
"source": [
"# import stuff\n",
"from TTS.config import load_config\n",
"from TTS.tts.datasets import load_tts_samples\n",
"from TTS.tts.utils.text.tokenizer import TTSTokenizer\n",
"from tqdm import tqdm\n",
"from matplotlib import pylab as plt\n",
"from multiprocessing import Pool, cpu_count\n",
"\n",
"# extra imports that might not be included in requirements.txt\n",
"import collections\n",
"import operator\n",
"\n",
"%matplotlib inline"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"Collapsed": "false",
"tags": []
},
"outputs": [],
"source": [
"# Load config.json properties\n",
"CONFIG = load_config(CONFIG_FILE)\n",
"\n",
"# Load some properties from config.json\n",
"CONFIG_METADATA = load_tts_samples(CONFIG.datasets)[0]\n",
"CONFIG_METADATA = CONFIG_METADATA\n",
"CONFIG_DATASET = CONFIG.datasets[0]\n",
"CONFIG_PHONEME_LANGUAGE = CONFIG.phoneme_language\n",
"CONFIG_TEXT_CLEANER = CONFIG.text_cleaner\n",
"CONFIG_ENABLE_EOS_BOS_CHARS = CONFIG.enable_eos_bos_chars\n",
"\n",
"# Will be printed on generated output graph\n",
"CONFIG_RUN_NAME = CONFIG.run_name\n",
"CONFIG_RUN_DESC = CONFIG.run_description\n",
"\n",
"# Needed to convert text to phonemes and phonemes to ids\n",
"tokenizer, config = TTSTokenizer.init_from_config(CONFIG)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"Collapsed": "false",
"tags": []
},
"outputs": [],
"source": [
"# print some debug information on loaded config values\n",
"print(\" > Run name: \" + CONFIG_RUN_NAME + \" (\" + CONFIG_RUN_DESC + \")\")\n",
"print(\" > Dataset files: \" + str(len(CONFIG_METADATA)))\n",
"print(\" > Phoneme language: \" + CONFIG_PHONEME_LANGUAGE)\n",
"print(\" > Used text cleaner: \" + CONFIG_TEXT_CLEANER)\n",
"print(\" > Enable eos bos chars: \" + str(CONFIG_ENABLE_EOS_BOS_CHARS))"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"def get_phoneme_from_sequence(text):\n",
" temp_list = []\n",
" if len(text[\"text\"]) > 0:\n",
" #temp_text = text[0].rstrip('\\n')\n",
" temp_text = text[\"text\"].rstrip('\\n')\n",
" for rm_bad_chars in CHARS_TO_REMOVE:\n",
" temp_text = temp_text.replace(rm_bad_chars,\"\")\n",
" seq = tokenizer.text_to_ids(temp_text)\n",
" text = tokenizer.ids_to_text(seq)\n",
" text = text.replace(\" \",\"\")\n",
" temp_list.append(text)\n",
" return temp_list"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"Collapsed": "false",
"tags": []
},
"outputs": [],
"source": [
"# Get phonemes from metadata\n",
"phonemes = []\n",
"\n",
"with Pool(cpu_count()-1) as p:\n",
" \n",
" phonemes = list(tqdm(p.imap(get_phoneme_from_sequence, CONFIG_METADATA), total=len(CONFIG_METADATA)))\n",
" phonemes = [i for sub in phonemes for i in sub]"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"Collapsed": "false",
"tags": []
},
"outputs": [],
"source": [
"s = \"\"\n",
"phonemeString = s.join(phonemes)\n",
"\n",
"d = {}\n",
"collections._count_elements(d, phonemeString)\n",
"sorted_d = dict(sorted(d.items(), key=operator.itemgetter(1),reverse=True))\n",
"\n",
"# remove useless keys\n",
"sorted_d.pop(' ', None)\n",
"sorted_d.pop('ˈ', None)\n",
"\n",
"phonemesSum = len(phonemeString.replace(\" \",\"\"))\n",
"\n",
"print(\"Dataset contains \" + str(len(sorted_d)) + \" different ipa phonemes.\")\n",
"print(\"Dataset consists of \" + str(phonemesSum) + \" phonemes\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"Collapsed": "false",
"tags": []
},
"outputs": [],
"source": [
"print(\"5 rarest phonemes\")\n",
"\n",
"rareList = dict(sorted(sorted_d.items(), key=operator.itemgetter(1), reverse=False)[:5])\n",
"for key, value in rareList.items():\n",
" print(key + \" --> \" + str(value) + \" occurrences\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"Collapsed": "false"
},
"outputs": [],
"source": [
"# create plot from analysis result\n",
"\n",
"x = []\n",
"y = []\n",
"\n",
"for key, value in sorted_d.items():\n",
" x.append(key)\n",
" y.append(value)\n",
"\n",
"plt.figure(figsize=(50,50))\n",
"plt.title(\"Phoneme coverage for \" + CONFIG_RUN_NAME + \" (\" + CONFIG_RUN_DESC + \")\", fontsize=50)\n",
"plt.xticks(fontsize=50)\n",
"plt.yticks(fontsize=50)\n",
"plt.barh(x,y, align='center', alpha=1.0)\n",
"plt.gca().invert_yaxis()\n",
"plt.ylabel('phoneme', fontsize=50)\n",
"plt.xlabel('occurrences', fontsize=50)\n",
"\n",
"for i, v in enumerate(y):\n",
" plt.text(v + 2, i - .2, str(v), fontsize=20)\n",
" plt.text(v + 2, i + .2, \"(\" + str(round(100/phonemesSum * v,2)) + \"%)\", fontsize=20)\n",
" \n",
" \n",
"plt.show()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"Collapsed": "false"
},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.9.12"
}
},
"nbformat": 4,
"nbformat_minor": 4
}
|